Statistics from Altmetric.com
Antidepressants continue to generate controversy as a treatment for children and adolescents. Amidst this controversy, rates of emotional disorders in young people are rising and the vast majority of children and adolescents with mental health disorders do not receive treatment.1 Mood disorders remain a leading cause of adolescent suicide. This editorial will review the recent meta-analysis of antidepressants by Locher et al2 and discuss the conclusions within the context of clinical practice and known flaws in the evidence base.
The review is an update on an earlier meta-analysis which also examined the effectiveness of selective serotonin reuptake inhibitor (SSRI) and serotonin norepinephrine reuptake inhibitor (SNRI) antidepressants compared with placebo across a range of disorders. This adds greatly to meta-analyses that have focused on one disorder only, allowing conclusions to be made about difference in efficacy across different indications. The current review included 36 trials, 6778 participants and broadly reported similar conclusions: antidepressants appear to be more beneficial for anxiety disorders (10 trials, g=0.56, 95% CI 0.40 to 0.72, P<0.001) and obsessive-compulsive disorder (8 trials, g=0.39, 95% CI 0.25 to 0.54, P<0.001), compared with depressive disorders (17 trials, g=0.20, 95% CI 0.13 to 0.27, P<0.001). The depression–anxiety difference appears to be driven by a greater placebo response in depressive disorders; conversely response to both antidepressants and (especially) placebo is lower for OCD. Adverse effects are reported more frequently in children and young people taking antidepressants compared with placebo.
The authors discuss the issue of high placebo response in depression, and harnessing components of the placebo response in designing a low-intensity intervention. Indeed, in our recent IMPACT study, we unexpectedly found that a low-intensity brief psychosocial intervention, focused on positive support, increasing hope and working towards goals, was as effective as cognitive-behavioural therapy (CBT). We agree that further research into this is necessary.
However, we would like to comment on some limitations of this study. First, while we welcome the separation by drug class, we think it would have been useful to separate also by individual drug. Both a recent network meta-analysis and Cochrane meta-analysis found fluoxetine to be more effective than other SSRIs for paediatric depression. Conversely meta-regression has shown that more serotonergic SSRIs are more effective for anxiety disorders.3 In addition, the review found one large randomised controlled trial (RCT) in post-traumatic stress disorder, which showed SSRIs to be slightly less effective than placebo; however, this study requires replication.
Second, although they state that effect sizes have been found to be greater in adolescents in other studies, they did not examine the effect of age; of note, the mean overall age of the sample was relatively young (12.9 years) and this was the case for all disorders. Since major depression peaks in adolescence, and antidepressants are more likely to be used in clinical practice in this group, it is essential that we have good clinical data according to age group to inform clinical practice.
Third, the authors acknowledge that comorbidity could explain the small effect size for depression, stating that this could be due to the lack of clear phenotype for depression. They also state what we think to be a larger problem stemming from comorbidity: the rates reported within the included trials are not representative of clinical populations, varying from 6% to 56%. These rates are considerably lower than those found in UK pragmatic trials of adolescent depression, as opposed to industry-sponsored trials with stringent exclusion criteria. For example, the large UK ADAPT4 and IMPACT5 trials, which had few exclusion criteria, reported, respectively, that 86% and 75% of adolescents had at least one comorbid disorder. Alongside the exclusion of adolescents with more severe depression and suicidality (not discussed by the authors), this makes it difficult to extrapolate results to patients we see in clinics. We therefore do not know the effectiveness of antidepressants within this group, and the relative risk/benefits. Meta-regressions in depressed adults have demonstrated that as severity of depression increases the effect of SSRI compared with placebo becomes more significant. If adolescents with more severe depression were included in studies, then effect sizes may be greater. We simply do not know the effects of comorbidity on treatment response in depression—while it may lower effect sizes as the phenotype is less pure, it may also increase effect sizes given that antidepressants work well for anxiety disorders and OCD, and treating these may improve depressive symptoms. Further investigation of the effects of comorbidity is needed.
Fourth, while the authors repeat previous findings that antidepressants do increase the risk of suicidality compared with placebos, they do not counter this by discussing the dangers of undertreating depression. The highest risk for suicide attempts in depressed adolescents is actually prior to the initiation of antidepressants,6 and there is a strong negative correlation between sales and prescriptions of antidepressants and suicide, both between regions and over time in the same regions. A review of 574 youth suicides reported that only 1.6% had been exposed to antidepressants,7 and the Utah youth suicide study reported that it was the lack of help-seeking that was implicated in completed suicides, and not the adverse effects of antidepressant treatment.8 Thus, while antidepressants do increase suicidality in a small number, they are more likely to reduce suicide risk (through their antidepressant action). A careful discussion balancing these risks is needed with adolescents and their families to enable informed consent.
Fifth, the authors only include one unpublished study. Other meta-analyses have included unpublished data obtained from drug companies. This omission may have inflated effect sizes.
Lastly, the authors do not review the risks/benefits of psychotherapy and potential adverse effects. The meta-analysis of psychological treatment by Weisz et al is cited, which reported similar effect sizes for anxiety and depression, and non-significant effects sizes for multiproblem treatment. However, psychological treatment trials have been exposed to considerably less scrutiny than antidepressant trials, have also been subject to stringent exclusion criteria and adverse effects are rarely studied. In addition, the TADS trial remains the only large comparison of fluoxetine to CBT where the rate of response to fluoxetine (61%) was nearly twice that of CBT (43%) at 12 weeks.
In summary, any decision to prescribe antidepressants to children and young people must be taken on an individual basis, collaboratively with them and their families. They must be informed of the potential risks and benefits of all forms of treatment, as well as the risks of not treating the disorder. In mild to moderate disorders, the National Institute for Health and Care Excellence (NICE) guidance does not advocate first-line prescribing of antidepressants as the risks may well outweigh the benefits. Recommended initial treatment after a period of ‘watchful waiting’ is up to 3 months of non-directive supportive therapy, group CBT or guided self-help. However, the current NICE guidance now states that combined treatment with antidepressants and psychological therapy may be considered as a first-line approach in moderate to severe adolescent depression. While access to specific psychological treatments remains difficult for many young people, clinicians will also have to weigh up with families the risks of no treatment while waiting for therapy and whether there is sufficient justification for starting antidepressants. New evidence from the IMPACT trial5 also suggests that a brief psychosocial intervention can be as effective as a specific psychological therapy, potentially increasing access to psychosocial support for young people as a stand-alone treatment and in conjunction with antidepressants.
We strongly advocate for the urgent funding of pragmatic, large RCTs which do not exclude young people who may potentially benefit the most from antidepressants. Psychological treatment trials also need to ensure that they have broad inclusion criteria, are representative of typical clinical populations and also implement rigorous evaluation of adverse effects. Studies need to directly compare antidepressants against psychological therapies. Only then can we truly have an informed debate about the relative risks and benefits of these treatments.
Contributors BD led on the paper. POW substantially contributed to a final draft with original input.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests BD has received a fee for a psychosocial treatment manual from Lundbeck. POW has provided consultancies for Lundbeck and Takeda Interpersonal psychotherapy supervisor.
Provenance and peer review Commissioned; internally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.